Ann. N.Y. Acad. Sci. ISSN 0077-8923
ANNALS OF THE NEW YORK ACADEMY OF SCIENCES
Issue: Barriers and Channels Formed by Tight Junction Proteins
Cingulin, paracingulin, and PLEKHA7: signaling and cytoskeletal adaptors at the apical junctional complex
Sandra Citi, Pamela Pulimeno, and Serge Paschoud
Department of Molecular Biology, University of Geneva, Geneva, Switzerland Address for correspondence: Sandra Citi, Department of Molecular Biology, 4 Boulevard d’Yvoy, 1211–4 Geneva, Switzerland. [email protected]
Cingulin, paracingulin, and PLEKHA7 are proteins localized in the cytoplasmic region of the apical junctional complex of vertebrate epithelial cells. Cingulin has been detected at tight junctions (TJs), whereas paracingulin has been detected at both TJs and adherens junctions (AJs) and PLEKHA7 has been detected at AJs. One function of cingulinandparacingulinistoregulatetheactivityofRhofamilyGTPasesatjunctionsthroughtheirdirectinteraction with guanidine exchange factors of RhoA and Rac1. Cingulin also contributes to the regulation of transcription of several genes in different types of cultured cells, in part through its ability to modulate RhoA activity. PLEKHA7, together with paracingulin, is part of a protein complex that links E-cadherin to the microtubule cytoskeleton at AJs. In this paper, we review the current knowledge about these proteins, including their discovery, the characterization of their expression, localization, structure, molecular interactions, and their roles in different developmental and disease model systems.
Keywords: cingulin; paracingulin; PLEKHA7; ZO-1; p120ctn; junctions
features of TJs, confirming that TJ proteins, together
Cingulin and paracingulin
with AJ proteins, can be part of atypical junctional
Cingulin was discovered as a Mr 140 kDa pro- structures.
- tein that copurified with myosin II from intesti-
- The sequence of cingulin, its biochemical behav-
nal epithelial cells and was localized in an apical ior, androtaryshadowingelectronmicroscopyshow circumferential “belt” in these cells1—hence the that the molecule exists as a parallel homodimer of name cingulin, from the Latin cingere (to form a two subunits, each comprising a globular head dobelt around). Immunoelectron microscopy demon- main, a coiled-coil “rod,” and a small globular tail strated that cingulin is localized at the cytoplasmic (Fig. 1).10,11 Although such domain organization is surface of tight junctions (TJs) in intestinal epithe- similar to that of myosin II, the sequence of cingulin lialcells.1,2 Furthermore,thesubcellularlocalization does not indicate either the presence of an actinand tissue distribution of cingulin indicate a close activated MgATPase activity or a propensity of the correlation with the continuous TJ belt of differen- coiled-coil rod to form higher order multimolecular tiated epithelia2–4 and endothelia,2,5 as well as the assemblies, such as myosin.11 Cingulin binds to and continuous and discontinuous junctions of strat- bundles actin filaments in vitro and also interacts ified epithelia.2,6,7 Interestingly, expression of cin- in vitro with myosin, suggesting that it could link gulin can be induced in fibroblasts by differentiating TJ proteins to the actin cytoskeleton.10,12 However, agents,resultinginitstargetingtospot-likeadherens since no major changes in the organization of the junctions (AJs).8 Moreover, cingulin is detectable at actin cytoskeleton have been detected in cingulinsites, such as the kidney glomerular slit diaphragms9 depleted cells, it seems that cingulin does not play and the outer limiting membrane of the retina,2 a significant role in controlling the architecture of which do not have the characteristic morphological actin filaments.
doi: 10.1111/j.1749-6632.2012.06506.x
ꢀc
Ann. N.Y. Acad. Sci. 1257 (2012) 125–132 2012 New York Academy of Sciences.
125
Cingulin, paracingulin, and PLEKHA7
Citi et al.
Figure 1. AmolecularmachinerylinkingTJstoAJs. Simplified diagram representing the molecular interactions between cingulin
(CGN), paracingulin (CGNL1), PLEKHA7, the TJ protein ZO-1, and the AJ protein p120ctn. The TJ and AJ domains of the plasma membrane are represented on the left, with schematics of JAM-A (purple) and claudins/occludin (blue) at TJs and E-cadherin (blue) at AJs. The major structural domains of the proteins are illustrated by colored boxes, whereas the regions involved in mutual interactions are highlighted by red rectangles, linked by lines with arrows. The head and rod domains are shown for CGN and CGNL1, with the ZIM (ZO-1–interaction motif) regions in the N-terminal part of the head,16,30 and the PLEKHA7-interacting region C-terminal to the ZIM in CGNL1.16 Dotted lines above the schemes of CGN and CGNL1 indicate the regions that interact with GEFs in vitro.21,22,27 ZO-1 contains PDZ domains that mediate the interaction with transmembrane proteins, Src homology 3 (SH3), guanylate kinase (GUK), and actin-binding-region (ABR) domains.33 The CGN- and CGNL1-binding region of ZO-1, as identified by yeast two-hybrid screens, is composed within the region C-terminal to the ABR. The CGNL1- and p120ctn-interacting regions of PLEKHA7 are distinct and located in the central region of the molecule. p120ctn contains a central region with armadillo (ARM) repeats. Afadin is another key component of the apical AJ belt but has been omitted from this scheme for the sake of clarity.
[Correction added July 17, 2012 after online publication: Figure has been changed to show the correct dotted lines above the scheme of CGN]
Paracingulin (Mr 150–160 kDa) is likely to be rescence shows junctional and apical localizations, a paralogue of cingulin—e.g., the two proteins whereas immunoelectron microscopy shows an exprobably arose from a gene duplication event clusive TJ localization. Furthermore, in intestinal and hence the name “para”-cingulin.13,14 Cingulin tissue, paracingulinisassociatedwithnonjunctional and paracingulin show a good degree of sequence actin filaments in the basal region of the cells.15 homology, especially in the coiled-coil rod do- A nonjunctional localization of paracingulin along main (39% identity) and in the N-terminal ZO-1– actin stress fibers has also been observed in fibrobinteracting motif (ZIM; 73% identity; Fig. 1).13,15,16 lasts expressing exogenous paracingulin.15,16 ThereThus, we speculate that paracingulin also exists as fore, although the actin- and myosin-binding propa parallel dimer. The head region of paracingulin erties of paracingulin are not known, paracingulin is larger than the head domain of cingulin (ap- may have a more direct association with the actin proximately 600 versus approximately 350 residues, cytoskeleton than cingulin.
- depending on species), and the coiled-coil rod is
- Unlike cingulin, paracingulin is structurally
smaller (approximately 660 residues versus approx- linked to the microtubule cytoskeleton. Perturbimately 800), accounting for the difference in their ing the organization of microtubules results in apparent molecular size. Paracingulin was inde- loss of junctional paracingulin, whereas cingulin pendently characterized as a junction-associated is unaffected.14 This is probably due to the intercoiled-coil protein (JACOP).15 Immunoelectron action of paracingulin with PLEKHA7,16 which is microscopy analysis showed that unlike cingulin, indirectly bound to microtubules17 (see later). In the localization of paracingulin is more promiscu- contrast, both cingulin and paracingulin localizaous. In kidney tissue, where paracingulin mRNA tions are dramatically affected by actin filamentis detected at high levels, paracingulin is localized disrupting drugs,14 confirming the association of both at TJs and AJs.15 In liver tissue, immunofluo- both proteins with the actin cytoskeleton. Indeed,
ꢀc
Ann. N.Y. Acad. Sci. 1257 (2012) 125–132 2012 New York Academy of Sciences.
126
Citi et al.
Cingulin, paracingulin, and PLEKHA7
the dynamics of junction exchange of cingulin and of cingulin does not have any detectable effect on paracingulin are similar to actin and more rapid junction assembly and development of the TJ barthan those of ZO-1.14 Cingulin and paracingulin rier in kidney cells, and the junctional targeting and can be detected in a complex together, and with activity of p114RhoGEF in these cells depends on other TJ proteins.14,16 However, since the junctional theFERMdomain–containingproteinLulu2, rather recruitment and dynamics of cingulin are indepen- than cingulin.24
- dent of paracingulin and vice-versa, the two pro-
- Significantly, the increase in cell proliferation,
teins likely do not function as a unit, but rather cell density at confluence, and claudin-2 expresas independent proteins, with partially redundant sion in cingulin-depleted kidney epithelial cells are
- functions.14
- rescued by the inhibition of RhoA, indicating that
In summary, cingulin and paracingulin are struc- RhoA mediates, at least in part, the effects of cinturally homologous proteins, with an asymmetric gulin on transcriptional and cell cycle regulation.22 shape, similar dynamics, partially overlapping sub- It is important to note that RhoA is modulated by cellular localizations, and distinct interactions with several other junctional molecules,25 and that tran-
- the actin and microtubule cytoskeletons.
- scription factors such as ZONAB and YAP are pu-
tative downstream effectors of RhoA.26 Therefore, the activity of cingulin as a RhoA regulator must be viewed in the context of a wider signaling network that is likely to exist in different cell-specific configurations.
Cingulin and paracingulin as adaptors for guanidine exchange factors
The first direct approach to study the function of cingulin was to generate embryonic stem (ES) cells lacking cingulin through homologous recombination. The results showed that embryoid bodies (EB) obtained by differentiation of cingulin knockout (KO) ES cells had apparently normal TJs, based on immunofluorescence, electron microscopy, and permeability assays, but had a remarkably altered pattern of gene expression: over 800 genes showed greater than twofold change in expression.18,19 Since these genes include several transcription factors, downstream target genes, and proteins such as claudin-2—whose expression is correlated to cell differentiation—their alteration indicates that cingulin is a regulator of signaling pathways that ultimately affects transcription and differentiation. Interestingly, the cingulin promoter has been identified as a target of HNF-␣, a transcription factor regarded as a key regulator of intestinal differentiation.20
To address the molecular mechanisms through which cingulin regulates gene expression and cell differentiation, we isolated cingulin-depleted kidney epithelial cell clones and discovered that another effect of cingulin depletion is an increase in RhoA activation in confluent monolayers, due to the junctional sequestration of the RhoA activator GEF-H1.21,22 Cingulin binds directly to GEF- H1 in vitro19,22 and has also been implicated in the junctional recruitment of a second Rho GEF, p114RhoGEF, which is important for junction assembly in corneal cells.23 However, depletion or KO
Paracingulin also interacts with the RhoA activator GEF-H1; consequently, depletion of paracingulin in confluent cells results in increased RhoA activity, an effect that phenocopies cingulin depletion. However, unlike cingulin, paracingulin also interacts with the Rac1 activator Tiam1. This results in a delay in the establishment of the TJ barrier, reduced peak of transepithelial resistance, and decreased Rac1 activation during the process of junction assembly/barrier formation in MDCK cells.27 Indeed, Tiam1 recruitment at junctions of MDCK cells is impaired in paracingulindepleted cells, indicating an important role of paracingulin in Rac1 activation via the junctional recruitment of Tiam1.27 Tiam1 has been shown to play a key role in Rac1 regulation in different cell model systems, to interact with the Par polarity complex, and to regulate Ras-induced oncogenesis.28,29
In summary, although it is not clear to what extent paracingulin is implicated in the regulation of gene expression and cell proliferation, through RhoA or other mechanisms, the available data indicate that cingulin and paracingulin contribute to fine-tuning the activity of Rho family GTPases both duringjunctionassembly(paracingulin)andatconfluence (cingulin and paracingulin). Furthermore, modulation of RhoA activity by cingulin is involved in its ability to control cell proliferation and gene expression.
ꢀc
Ann. N.Y. Acad. Sci. 1257 (2012) 125–132 2012 New York Academy of Sciences.
127
Cingulin, paracingulin, and PLEKHA7
Citi et al.
Identification of PLEKHA7 as a paracingulin-interacting protein
creased expression of paracingulin, indicating that PLEKHA7 plays a major role in the recruitment and stabilization of paracingulin to AJ. Indeed, the association of paracingulin with PLEKHA7 at AJs can explain why ZO-1 depletion only delays, but does not significantly decrease, paracingulin junctional recruitment.16
The mechanisms controlling subcellular localization of paracingulin should be investigated in additional cell types, considering the heterogeneous localization of paracingulin in tissues. For example, it is not clear which molecular interactions target paracingulin to actin stress fibers in epithelial and nonepithelial cells.15,16 This targeting occurs independently of the interaction with ZO-1– and PLEKHA7, since it is observed in Rat-1 fibroblasts using exogenous mutant constructs lacking the ZO-1– and PLEKHA7-interacting regions.16 The functional role of paracingulin at stress fibers remains unknown, but considering its ability to interact with GEFs that regulate Rho family GTPases, a putative role in regulating RhoA and Rac1 activities at this site could be hypothesized.
Cingulin and paracingulin are both localized at TJs, butparacingulinisalsodetectedatAJsandinassociation with extra-junctional actin filaments. What is the molecular basis for these different localizations? Recent studies have provided some answers to these questions. ZO-1 contributes to TJ recruitment of both cingulin and paracingulin, as indicated in experiments showing that in cells depleted of ZO-1, but not ZO-2, the junctional recruitment of cingulin is decreased and the junctional recruitment of paracingulin is delayed.16 Both proteins bind to ZO-1 in vitro through their conserved ZIM in the N-terminal region of the globular head domain (Fig. 1), and deletion of the same region abolishes cingulinjunctionaltargetingintransfectedcells.16,30 Thus, it is reasonable to conclude that cingulin and paracingulin directly bind to ZO-1 at TJs. However, despite the observation that in an epithelial breast cancer line ZO-1 KO completely abolishes cingulin junctional localization,31 a normal junctional localization of cingulin was observed in tissues from ZO-1 KO mouse embryos.32 Therefore, the mechanisms of cingulin junctional recruitment appear to be cell-context dependent, and redundant interactions can target cingulin, and possibly also paracingulin, to TJs. Indeed, in vitro cingulin binds not only to ZO-1 and ZO-2 but also to ZO-3 and other TJ proteins.10 Interestingly, the region of ZO-1 suffi- cient for interaction with either cingulin or paracingulin (by the yeast two-hybrid technique) is the C-terminal region adjacent to the actin-binding region (Fig. 1). Thus, the C-terminal region of ZO-1 appears to be a structural module dedicated to interaction with actin, actin-binding proteins, and other cytoskeletal proteins.33
PLEKHA7: a new component of the apical AJ belt
PLEKHA7 was discovered independently by Takeichi’s and our laboratory as a novel protein component of epithelial AJs.16,17,34 The distinguishing feature of PLEKHA7, with respect to most other AJ proteins, except afadin, is that it is confined to a continuous apical belt and is absent from the lateral membrane.34 In the pancreas, for example, p120ctn and PLEKHA7 labeling are only partially overlapping, with PLEKHA7 strongly labeling ducts, whereas p120ctn is localized along both junctional and lateral surfaces of exocrine acinar cells (Fig. 2). The localization of PLEKHA7 in several tissues is highly reminiscent of the localization of TJ proteins and similar to that of afadin,34 another AJ protein that has a belt-like distribution and has alternatively been described as a TJ protein.35,36 Double labeling of PLEKHA7 with ZO-1 shows a partial co-localization, which may depend on tissue type and section orientation (Fig. 2). However, the overall localization and tissue distribution of ZO-1 and PLEKHA7 are not identical, and immunoelectron microscopy of intestinal cells shows a concentration of PLEKHA7 labeling at AJs.34
A yeast two-hybrid screen revealed PLEKHA7 as a very high-confidence interactor of paracingulin; and we showed that the AJ targeting of paracingulin in MDCK cells is mediated by its interaction with PLEKHA7.16 PLEKHA7 is an AJ protein34 comprising WW, proline-rich, PH, and coiled-coil domains (Fig. 1). An N-terminal region of the head domain of paracingulin, distinct from the ZO-1–interacting region, interacts with a region in the C-terminal domain of PLEKHA7, the first coiled-coil domain16 (Fig. 1). Depletion of PLEKHA7 in kidney epithelial cells results in a loss of junctional staining and de-
ꢀc
Ann. N.Y. Acad. Sci. 1257 (2012) 125–132 2012 New York Academy of Sciences.
128
Citi et al.
Cingulin, paracingulin, and PLEKHA7
membrane domain, results in the redistribution of both p120ctn and PLEKHA7, indicating that PLEKHA7 may also be associated with N-cadherin based junctions, through p120ctn.17 However, the observation that PLEKHA7 is only associated with p120ctn along the AJ belt, but not along the lateral surfaces of polarized epithelial cells (Fig. 2),34 suggests that additional interactions are essential to target PLEKHA7 to the junctional belt.
Cingulin and PLEKHA7 in development and disease
Insights about the functions of junctional proteins can be obtained from studies about their roles in development and disease. Cingulin is maternally expressedandcorticallylocalizedbothinmouseandin Xenopus laevis embryos.37–40 In the mouse early embryo, cytocortical maternal cingulin is degraded by endocytic turnover, and zygotically expressed cingulin accumulates at TJs after the 16-cell stage, just after ZO-1.37,38 Moreover, cingulin is upregulated in the mouse blastocyst trophectoderm, and its stability requires cell–cell contact.38 In the frog embryo, maternal cingulin is recruited from the apical cortex into the newly forming TJs starting from the first cleavage, and defines the border between apical and lateral membranes.39,40 Depletion or KO of cingulin from either Xenopus embryos or mice, using either morpholino oligos or gene targeting techniques, respectively, does not impair embryo viability (our unpublished results). However, it was recently reported that morpholino-mediated depletion of cingulin from the developing neural crest of the chick midbrain expands the size of the migratory neural crest cell domain.41 Moreover, in the chicken embryo, overexpression of cingulin is correlated with changes in RhoA protein distribution in ventrolateral neuroepithelial cells, although it is not clear whether RhoA activity is altered in this model.41 In-
Figure 2. The localization of PLEKHA7 at the apical AJ belt.
Double immunofluorescent localization of PLEKHA7 (red) and p120ctn (green) in frozen sections of pancreas (top), and of ZO- 1 (red) and PLEKHA7 (green) in frozen sections of duodenum. Samples were obtained and processed as described.34 The boxed areas in the low magnification images are shown enlarged below. Arrowheads indicate lack of co-localization, and arrows indicate co-localization. Nuclei are labeled in blue by DAPI in merged images. Bar = 2 µm. Note that PLEKHA7 is not co-localized with p120ctn along the lateral surface of acinar epithelial cells (see also Ref. 34) but co-localizes with p120ctn along intralobular and interlobular ducts. ZO-1 is not highly expressed in the duodenum, but in some sections of cells expressing sufficient amounts of ZO-1, there is a partial co-localization of ZO-1 with PLEKHA7, indicating that the apical, PLEKHA7-containing AJ belt is immediately adjacent to the ZO-1–containing TJ belt.